In common practice, quay walls are engineered to fulfil their services for a minimum lifetime. However, in many cases, quay walls are designed stronger than minimally required, which means that the quay wall could be able to serve its purpose over a longer period or endure higher loads. In general, the optimization of quay wall designs or the assessment of the extension load-bearing capabilities, particularly in existing structures, is substantiated through probabilistic design methodologies.

This study investigates the operation of reliability updating with deformation data to enhance the operational lifespan and load-bearing capacity of existing quay walls. Traditionally, quay walls are designed with predetermined probabilities of failure based on preliminary information gathered through desk studies and on-site investigations. Observations of existing structures provide new information on the actual deformation behaviour. This study acknowledges the conservative assumptions in initial designs, potentially resulting in concealed geotechnical and structural capacities. It introduces a novel approach involving reliability updating using monitoring data from 'smart' quay walls equipped with advanced sensors. These sensors, measuring especially lateral deformations, provide input for the updating process in a Bayesian model.

Application of this methodology is demonstrated through a theoretical and an actual case study, including the Sif quay wall in the Port of Rotterdam (Maasvlakte II).

The theoretical case study validated the methodology. This was done by updating two initial (prior) distributions for the friction angles of two soil layers using the observed maximum deformation for a combination of predefined values of the friction angles. The updated (posterior) distributions converged towards the predefined values and the distributions reduced in variation or became more informed. These findings affirmed the overall operation of the intended method and proved the efficiency and accuracy of the addition of metamodelling. However, the success of the metamodel appeared to strongly rely on appropriate settings for the specific situation.

For the case study of the Sif quay wall, its deformation behaviour was analysed for three different load cases accounting for the effect of excavation, the effect of water level fluctuations and the effect of top loading. The most recent deformation measurements, together with the knowledge of past deformation behaviour, have proved most useful for updating as all aspects contributing to the lateral deformation of the quay were present. In the updating process of the Sif quay wall case study, six stochastic variables were analysed: the strength (friction angles) and stiffness parameters (stiffness moduli) of three influential soil layers. Within the scope of this study, an increase in one or both parameter types means that the structure may have a higher functional capacity than expected, in the form of moment or normal force capacity.

Here, the strength parameters proved more prone to updating than the stiffness parameters and showed a significant increment. The initial parameter uncertainty indicated by the standard deviation was found to have minimal impact, but larger discrepancies between predicted and observed deformations led to more significant updates.
The quay wall at the Sif terminal appeared to have a negligibly small failure probability, even before updating. Therefore, the impact of updating was not measurable in terms of the reliability index, but the outcomes of the limit state function distribution were significantly impacted: their variation was reduced, and the overall distribution moved away from the point of failure.

Regarding cost-effectiveness, the risk mitigation for existing walls after monitoring and reliability updating might be limited. However, the potential for significant steel reduction in new quay wall designs based on updated soil information highlights the value of reliability updating. The approach of this study may lead to more efficient and sustainable quay wall construction practices.

The study also acknowledges limitations in the simplification of the soil model compared to the actual situation and the consideration of one single failure mechanism. Another important factor is the limited level of knowledge on the quay wall deformation conditions, such as loading magnitudes and potential time-dependent effects, compared to the minimized model uncertainties. This aspect should be taken into account when evaluating the results of the update results. Recommendations for further research include:
• To achieve a more realistic representation of quay wall behaviour, it is recommended to incorporate various failure mechanisms, including geotechnical failure, as real-world failures often involve the interplay of multiple mechanisms.
• For a more robust prediction, the model should explore a wider range of stochastic variables, including factors like bed level, water level fluctuations, and top load.
• To enhance the accuracy of the updated parameters, it is recommended to acquire additional measurement data through on-site loading tests or by implementing smart bollards for real-time loading information.
• Investigating the correlation between the stochastic variables is crucial for understanding their combined influence on the model's predictions. This analysis will lead to more accurate results.
• To maximize its practical value, the model should be extended for applications such as berth deepening or designing for increased top loads.

In summary, this research highlights the potential of reliability updating with deformation data, offering more insights into a possible increase of a quay wall's capacity. By understanding a quay wall's reliability, port authorities can optimize its use and extend its service life, reducing environmental impact and bringing economic benefits.